Three-year Population Health Project Meets Data Milestone

Southeastern Diabetes Initiative (SEDI) closed database on June 30, 2016

July 18, 2016
The Southeastern Diabetes Initiative (SEDI), a three-year population health project focused on adults living with Type 2 Diabetes Mellitus (T2DM), closed on June 30, marking the culmination of patient care, data collection, and data cleaning for four study sites.

With all data collection complete and the database locked and archived, the Center for Predictive Medicine statisticians have initiated their final analysis of the study outcomes.

The SEDI project team, initiated by original PI Rob Califf and now led by PI Ebony Boulware after his transfer to the U.S. Food and Drug Administration, celebrated this major achievement under the operational leadership of PL Brooke Heidenfelder (pictured above, center, with Lisa Davis [right], and Shelley Rusincovitch [left]). Key collaborating groups at Duke include the DTMI, the DTRI, the DCRI (Center for Predictive Medicine, Center for Data Management, and the Center for Pragmatic Health Services Research), and various departments within the School of Medicine.

"This was a population health study, different from the usual type of study that the DCRI manages," said Heidenfelder. It includes four innovative elements:

  • The use of an algorithm (developed by Heidenfelder's team) to stratify patients into low, moderate, or high risk categories;
  • The use of Community Health Workers (CHWs) in care delivery design;
  • The integration of geospatial data;
  • The collection of electronic health record data from collaborating health systems and combining them into the SEDI Datamart.

"This study is very different from the work the DCRI usually does because there is no drug or device involved.  We have gleaned great experience in running this population health study and determining how many people have to be involved at the community level.  Through this study we are trying to move the needle on diabetes by looking at a spectrum of factors, including mental health, poverty, transportation, and food deserts," Heidenfelder said. "It provides a multidimensional understanding of individual and community health status and vulnerabilities."

The 2013 Health Affairs article co-written by Califf notes that "While the growth of electronic health data is rapidly growing, most data systems don't connect individual patient records to the data sets from outside the health care delivery systems. These isolated data systems can't support efforts to recognize or address how the physical and environmental context of each patient influences health choices and health outcomes." The SEDI study aimed to collect geographic health information to provide a comprehensive information base (and template) for community health assessment and intervention.

The three-year intervention included a clinical intervention for high-risk patients, a telephone module intervention for moderate to low risk patients, and a community intervention for low risk patients.

Total enrollment was 648 among four study sites, including Durham County and Cabarrus County, N.C.; Quitman County, Miss.; and Mingo County, W.V.  Each site varied in population size and had its own characteristics: rural vs. urban, demographic makeup, socioeconomic status, and number of Spanish speakers. Each site also had a community advisory board, and determined what roles to include in their clinical teams in addition to a provider and a CHW.

According to Heidenfelder, West Virginia and Mississippi sites generally were not as affluent as the North Carolina sites, which influenced the obstacles faced by patients in each area. Mississippi was incredibly rural and participants dealt with transportation issues. West Virginia was a tight-knit community and not always receptive to outsiders, which meant the collaboration with the local diabetes coalition was that much more valuable.

There was excellent collaboration between the clinical teams and local organizations. For example, in Durham, Healing with CAARE, Inc. allowed a diabetes support group to form and the addition of a special diabetes food pantry, both of which continue to exist even after the study concluded.

The Community Health Workers (CHW) formed trusted bonds with the participants. "They are empathetic and knew about the resources available to help people with various needs, including completing applications, help with filling prescriptions, and finding transportation. When you see some of the barriers that patients face, you really want to help them, and the CHWs made a huge and positive impact," Heidenfelder said.

Funded by the Centers for Medicare & Medicaid Services Health Care Innovations Awards, and a grant from the Bristol-Myers Squibb Foundation, the project was designed to augment the existing standard of care in an effort to:

  • Improve population-level diabetes management, health outcomes, and quality of life for diagnosed and undiagnosed adults living with T2DM;
  • Reduce disparities in diabetes management, health outcomes and quality of life for adults living with T2DM (including disparities based on race, socioeconomic and insurance status);
  • Reduce healthcare costs associated with T2DM.

Heidenfelder hopes that between the datamart and the clinical data, the study will show that having a team that involves CHWs reduces costs by keeping people out of the hospital and/or emergency rooms, thus saving patients time, trouble, and health issues, and saving Medicare and Medicaid money.